Skip to main content
idapt
HomeCodeAI ModelsPricing
Sign inStart free trial
  • Home
  • Pricing
  • AI Models
  • Image models
  • Voice models
  • Video models
  • Rankings
  • New models
  • Model status
  • Multi-Model Chat
  • Agents
  • Computers
  • Drive
  • Automations
  • AI Gateway
  • All features →
  • LLM cost calculator
  • Token counter
  • All free tools →
  • Blog
  • Use cases
  • Comparisons
  • Best of
  • Benchmarks
  • Changelog
  • Help center
  • FAQ
  • Privacy
  • Compare all models
  • Support
  • idapt Code
  • Developers
  • Quickstarts
  • API reference
  • API pricing
  • CLI
  • MCP
  • Downloads
  • Desktop
  • Badges and embeds
© idapt[email protected]TermsPrivacy PolicyLegal noticeReport content
X (Twitter)
Help Center
🧠

Agent memory

Found this helpful? Share it:

Memory lets your agents remember things across conversations. An agent writes down what matters as it works and reads it back the next time you talk, so you don't have to re-explain the same context every time 🧠

How agent memory works

Every agent keeps its own memory: a private set of notes that lives outside any single chat. At the start of a conversation the agent reads its memory, and as it works it writes durable facts back.

Because memory lives outside the chat, it survives the conversation. Close the tab, come back a week later, start a brand-new chat. The agent still remembers what it learned.

Memory is for durable facts, like your preferences or a project's details, not a transcript of every message. The agent decides what is worth keeping and updates its memory as things change.

Each agent's memory is its own

Each agent you create gets its own memory automatically, and it belongs to that agent alone. A research agent and a support agent never mix up what they know, and one agent cannot read another's memory.

An agent's memory is private to that agent and to the people who can manage it. It is not shared with everyone in your workspace, and it does not show up in the workspace search.

Turn memory off for a chat

Sometimes you want a clean slate: a one-off conversation where the agent neither reads its memory nor writes anything new. You can switch memory off for a single chat without touching the agent's saved memory.

  1. Open the tools dropdown in the composer, the sliders icon next to where you type.

  2. Toggle Memory off. A check mark means it is on; an × means it is off for this chat.

  3. Send your message. The agent answers without touching its memory.

The toggle is per chat, so it only affects the conversation you are in. Your other chats with the same agent keep memory on.

Turn memory on or off for an agent

Each agent has a memory switch in its settings. Turn it off and the agent stops using memory entirely. Turn it back on and the same memory comes back, exactly as it was, so toggling it off does not erase anything.

Next to the switch is an Open memory button. Open it to read every note the agent has saved, edit a fact that is wrong, or delete something stale.

Memory has its own size limit, separate from your Drive storage. What an agent remembers counts toward that limit, not toward your files.

Use memory from outside idapt

An agent's memory is available beyond the app. Through the idapt API, SDK, CLI, and MCP, an external agent can remember and recall against the same memory. That lets a coding agent in your terminal share what it learns across sessions. See Connect any coding agent.

FAQs

Related articles

🤖

Agents

Create AI personas with custom instructions, their own memory, and their own default model.

🔌

Connect any coding agent

Point any coding agent or MCP client at your idapt workspace over the CLI or MCP, then use idapt Memory and act as an agent.

Was this helpful?